RDP 8606: The Nation's Wealth – Some New Calculations for Australia 2. Concepts, Methods and Earlier Estimates of Australia's wealth
June 1986
(a) Some Definitional Issues
It is first necessary to say what is meant by personal wealth. It consists of physical assets, such as houses and consumer durables, and of claims on other sectors net of liabilities to other sectors. This raises the question of the definitions of assets and liabilities to be included. Truly comprehensive valuation is impractical because of data and conceptual difficulties. Important exclusions are human wealth and “social property”, or the rights to benefits from the state – access to communal assets such as government schools and state pension rights for example.[4]
A question related to the definition of wealth is the basis for valuation. Two well known possibilities exist: the “realisation” basis (or the value obtained in a sale on the open market at the date in question) and the “going concern” basis (or the value to a person or household on the assumption that the asset is retained). These valuations may be different for at least two reasons. First, the value of an asset to an intra-marginal holder may exceed its market price, even in a perfect market. Second, markets may not be perfect: an example is occupational pension rights, where realisation value may be zero. In this study, market realisation value has been used: where this is not available an approximation has been employed.
The literature on wealth stock estimation, and particularly on the development of sectoral balance sheets, draws a distinction between personal and private wealth. Personal wealth is thought of as the value of assets to which households in some sense have title. Private wealth, by contrast, encompasses all non-public assets. In a simple economy, the two magnitudes would be equal for any given valuation convention. In practice, however, there are at least two differences, leaving aside (for the moment) the question of foreign ownership. Firstly, the wealth of non-profit organisations, such as schools and churches, is private but not personal. Secondly, the difference between the net value of assets of listed corporations and their stock exchange valuation is private, but is not included in personal wealth, since equity holders have title to only the stock exchange valuation of their shares. The wealth estimates presented here do not distinguish between private and personal wealth. The difference between the two magnitudes is not considered to be quantitatively significant when compared with the aggregate value of wealth under either definition, and the terms will be used interchangeably throughout the paper.
(b) Alternative Approaches to Wealth Stock Estimation
Methods of aggregate wealth estimation can be conveniently split into the estate, survey and inventory approaches. The estate method, which uses information on deceased estates and age-specific mortality statistics to estimate both aggregate personal wealth and its distribution, is perhaps the most famous.[5] It has been used for well over a century;[6] Australian studies using this method include Gunton (1971, 1975); Raskall (1977); and Knibbs (1918). Of these, only Knibbs was primarily interested in aggregate wealth assessment, as opposed to wealth distribution. A feature of the estate method is that some wealth will be missing from the aggregate figure, for three reasons. First, the property of wealth – holders excluded from the estate data because their valuation below the dutiable threshold is omitted; second, certain assets are not appropriately valued, and are usually assessed below market value; third, certain types of property may be missing from estimates recorded in the statistics.
The second approach, the survey, may also be expected to exclude some part of aggregate personal wealth, because of both under-response and under-reporting, especially by the relatively wealthy.[7]
The inventory approach, by contrast, relies on independent information about the total holdings of particular assets. It amounts to summing the values of privately owned physical and/or financial assets, avoiding problems of double-counting. Adjustments must be made for public debt held by the private sector, and for net foreign asset holdings. There is no reason to suppose that an inventory estimate will be systematically biased, provided that all assets are covered and that they are appropriately valued. The inventory approach has the additional advantage that it can exploit independent estimates of wealth components which have been developed for administrative or policy purposes.
(c) Previous Estimates of Australia's Current Wealth
A number of estimates of total Australian wealth since the 1960s now exist, of which the best known are perhaps those of Helliwell and Boxall (H-B) (1978) and Williams (1983). Most employ some form of inventory approach. Table 1 reports aggregate wealth from a number of these studies, for selected years from 1953 to 1981. Several points deserve comment. Firstly, estimates reported in the first four rows of the table are closely related to one another. The Norton et al. (1982) estimates are revisions and updates of the H-B estimates. More recent unpublished Reserve Bank revisions to this series have increased the June 1981 estimate to $295 billion. The most recent Reserve Bank figures are used in Section 3. In all these estimates, land valuations (both urban and rural) were taken from the work of Scott (1968), who used local government assessments as a basis for estimation.[8] These are not regarded as reliable guides to market valuation, and estimates of wealth are reported both including and excluding land values. The dwelling stock (excluding land) was valued at replacement cost.
| 1953 | 1959 | 1964 | 1968 | 1970 | 1975 | 1981 | |
|---|---|---|---|---|---|---|---|
| Helliwell/Boxall (1978) (Table 3, pp.59–60) (1) |
– | 34.2 | 46.1 | 65.1 | 64.8 | 107.1 | – |
| (2) | – | 39.9 | 56.2 | 80.1 | – | – | – |
| Norton/Garmston/ Brodie (1982) (Table 6.10, p.168) (1) |
– | 34.2 | 46.8 | 66.5 | 66.5 | 114.5 | 230.5 |
| (2) | – | 40.1 | 56.9 | 81.4 | – | – | – |
| Williams (1983) (Table 3, p.58) |
– | – | – | 63.9 | 78.6 | 165.5 | 360.5 |
| Raskall (1977) (Table 6, p.25) |
– | – | – | – | 85.7 | – | – |
| Gunton (1975) (Table 6.1, p.119) |
18.7 | 26.6 | 38.6 | 55.9 | 61.9 | – | – |
| Podder/Kakwani (1973) (Derived by Raskall (1977), p.8) |
– | – | – | 50.2 | – | – | – |
1. Excluding value of land |
|||||||
An important element of these estimates of total private wealth is the value of business assets and inventories. The H-B studies employ an industry assets approach, using share market and balance sheet data for a sample of firms to infer market valuations of business assets and inventories for the whole economy. This includes the non-land capital stock of farms, an imputation which is especially questionable. Adjustments are then made for intersectoral claims and liabilities.
The H-B estimates, then, while of a pioneering kind, can be criticised on a number of grounds. In my view, the following deficiencies are the most important. First, no attempt was made to place a market value on land or housing, even though both residential housing and farms are subject to considerable market volatility, and both are important components of overall wealth. Secondly, agriculture is so different from the bulk of incorporated activity, both in its experience of output price behaviour and in its production techniques, that valuation of non-land capital in Agriculture by appeal to share market data is unlikely to provide reliable figures. Thirdly, the sample used in generating the market valuation ratio for business fixed assets and inventories covered only 65 firms, which ten years ago accounted for no more than 20 per cent of incorporated business activity. Further, book valuation of assets is a dubious multiplier in moving from the sample to an aggregate estimate, since such valuation will be systematically influenced by accounting practice and, in periods of inflation, asset life. Both may vary between the incorporated and unincorporated sectors. Fourthly, H-B omitted the debt issue of local and semi-government authorities from their estimate of government bonds, and this convention has been continued by the Bank in its updates. The value of government bonds is thus seriously underestimated.
By contrast, Williams (1983) relies on an inventory of the financial assets and liabilities of households, making use of data first constructed by Anstie et al (1983) on personal holdings of equities and reserves of pension funds and life offices. In choosing to use personal holdings of financial assets to value personal wealth other than housing and consumer durables, Williams explicitly excludes the value of unincorporated business fixed assets and inventories, including rural land. He does, however, develop housing stock (including land) valuation estimates, which have been derived from sales data, and these valuations typically comprise more than half of the value of his estimates of total personal wealth. The only other study to estimate the market value of residential land and housing is reported in an unpublished paper by Roger (1981).
Raskall (1977, pp.16–25) provides a further example of the inventory approach. His estimate is at first sight encouragingly close to the total value that might be anticipated if the land-inclusive valuations of H-B were projected forward to 1970. On the other hand, Raskall's valuation of residential land and buildings ($31 billion) is only two-thirds of Williams' ($46 billion) for the same year; unlike Williams, Raskall does not use direct market valuation in deriving his estimate.
In summary, among those studies which use an inventory type approach, agreement on aggregate estimates is patchy. Other approaches yield significantly lower values. The estimates reported by Gunton, who uses the estate method, reflect the “missing wealth” problem referred to above. Similarly, under-response and under-reporting can be inferred from Raskall's calculation of the value of total wealth implied by the Macquarie Survey of Consumer Finances and Expenditures data used by Podder and Kakwani (1973).
Footnotes
If communal assets such as government schools were added in there would be some double counting in that government bonds and Reserve Bank liabilities included in private wealth are used for deficit financing. Unofficial ABS public sector capital stock estimates are available in Walters and Dippelsman (1986). [4]
The method is outlined in Piggott (1984); for a detailed study see Atkinson and Harrison (1978). [5]
The first attempt of which I am aware is that of Baxter (1869), for the U.K. [6]
Another approach, the “investment income” approach, has also been used from time to time, especially in the UK. It is not discussed here, but an outline of the approach is provided in Piggott (1984). Atkinson and Harrison (1978) offer a detailed account. [7]
Scott's estimates were later updated to 1972 by Parkin et al (1975). [8]